HAL Id: hal-02156624
https://hal.archives-ouvertes.fr/hal-02156624
Submitted on 14 Jun 2019
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of
sci-entific research documents, whether they are
pub-lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Infrared spectroscopy as a useful tool to predict land use
depending on Mediterranean contrasted climate
conditions: A case study on soils from olive-orchards
and forests
Ninon Delcourt, Catherine Rébufa, Nathalie Dupuy, Nathalie Boukhdoud,
Caroline Brunel, Juliet Abadie, Isabelle Giffard, Anne Marie Farnet da Silva
To cite this version:
Ninon Delcourt, Catherine Rébufa, Nathalie Dupuy, Nathalie Boukhdoud, Caroline Brunel, et al..
In-frared spectroscopy as a useful tool to predict land use depending on Mediterranean contrasted climate
conditions: A case study on soils from olive-orchards and forests. Science of the Total Environment,
Elsevier, 2019, 686, pp.179-190. �10.1016/j.scitotenv.2019.05.240�. �hal-02156624�
Infrared spectroscopy as a useful tool to predict land use depending on
Mediterranean contrasted climate conditions: A case study on soils from
olive-orchards and forests
Ninon Delcourt
a, Catherine Rébufa
a, Nathalie Dupuy
a, Nathalie Boukhdoud
a, Caroline Brunel
a,
Juliet Abadie
a,b, Isabelle Giffard
a, Anne Marie Farnet-Da Silva
a,⁎
a
Aix Marseille Univ, Avignon Université, CNRS, IRD, IMBE, Marseille, France
b
UR RECOVER IRSTEA, Le Tholonet, 13100 Aix en Provence, France
H I G H L I G H T S
• FTIR-ATR was used to test land use ef-fect on soil chemical properties. • Soil chemical signature differs between
olive-tree orchards and Mediterranean forests.
• Climate (Humid vs Sub humid, coastal vs inland area) also shapes soil proper-ties.
• PLS modeling confirmed that FTIR-ATR is a useful tool to predict land use.
G R A P H I C A L A B S T R A C T
a b s t r a c t
a r t i c l e i n f o
Article history:
Received 25 February 2019 Received in revised form 16 May 2019 Accepted 17 May 2019
Available online 22 May 2019
Editor: Shuzhen Zhang
Soil chemical properties depend on various environmental factors such as above ground vegetation, climate and the parent rock substratum. Land use, and the associated management practices, is one of the major drivers which can deeply impact soil properties. To better understand the dynamics of soil chemical properties and to assess potential impact of land use, an improved monitoring of chemical signature in organo-mineral topsoils is necessary. Here, we explored how land use (forests or agrosystems i.e. olive-tree orchards) may shape soil chemical signature and whether it depends i) on the type of agricultural or sylvicultural practices, ii) on contrasted Mediterranean climate conditions at different spatial scales. We measured variations in soils proper-ties by FTIR-ATR (Fourier-Transformed Infrared– Attenuated Total Reflectance) spectroscopy and elemental con-centrations. FTIR showed that the aromatic fraction of organic matter and CaCO3discriminated soils under
different land uses (orchards or forests) and this depended on climate (sub-humid vs humid climate). Moreover, the chemical signatures of soils varied with the practices applied. For agrosystems, soils complemented with olive-mill wastes were characterized by aromatics compared to soils under natural grass or tillage. For forests, soils from Pinus spp. stands and Quercus spp. stands were discriminated by CaCO3and aromatics respectively.
Contrasted climate conditions at local scale, i.e. northern vs southern slopes for forests and distance from the sea (coastal vs inland area) for agrosystems, had an effect on soil chemical signature. The AcomDIM interpreta-tion of FTIR-ATR signals showed that factors“land use”, “practices” and “climate” and their interactions could
Keywords: AcomDim Agricultural practices Land use FTIR-ATR Mediterranean soils PLS ⁎ Corresponding author.
E-mail address:anne-marie.farnet@imbe.fr(A.M. Farnet-Da Silva).
https://doi.org/10.1016/j.scitotenv.2019.05.240
0048-9697/© 2019 Elsevier B.V. All rights reserved.
Contents lists available atScienceDirect
Science of the Total Environment
have a significant impact on soil chemical signature. PLS modeling also confirmed that FTIR-ATR is a useful tool to predict a type of land use depending on climate.
© 2019 Elsevier B.V. All rights reserved.
1. Introduction
Soil organic matter (SOM) is composed of a great diversity of mole-cules from plant, animal and microorganisms, at different stages of decay, which defines soil chemical signature. This chemical fingerprint depends on various environmental biogeophysical drivers and their in-teractions (Coûteaux et al., 1995,Garcia-Palacios et al., 2016,Liu et al., 2019), mainly climate, above ground vegetation and the parent rock substratum. Soil chemical signature thus results from a combination of different processes linked to organic matter transformation i.e. degrada-tion/stabilisation and their balance, which is subrogated to complex in-teractions of biological, chemical and physical processes in soils. The persistence and decomposition of SOM is known to be also influenced by the mineral fraction of soil and the proportion of the different com-ponents of this fraction depends on soil physico-chemical environment i.e. pH, redox potentials, temperature, humidity and soil parent material
(Kleber et al., 2015). The physical and mineral composition of soil also
contributes to its chemical signature as revealed for instance by spectro-scopic and X-Ray diffraction techniques (Prandel et al., 2014).
Climate is one of the major factors controlling these processes. In the Mediterranean area, climate conditions are particularly drastic i.e. char-acterized by intense summer droughts and high temperatures. This stress linked to water potential is reinforced in coastal environments by wind regime and osmotic stress via sea spray exposure, hampering decomposition processes (Farnet et al., 2016;Qasemian et al., 2014).
Land use is also expected to have a strong effect on the quality and composition of SOM (Dhillon et al., 2017). In forests, vegetation assem-blages are known to deeply impact SOM composition (Aranda et al.,
2011,Brunel et al., 2017,Prescott and Grayston, 2013.). Broadleaved
species, such as Quercus spp., enhance SOM quality, increasing easily-degradable compounds and thus enhancing microbial metabolism
(Polyakova and Billor, 2007). Conversely, some vegetal species specific
to the Mediterranean area such as Pinus spp. are particularly sclerophyllous (high amounts of recalcitrant molecules such as cutin or lignin) and poor in nutrients, which induces lower rates of decompo-sition (Iovieno et al., 2010). Agriculture and associated practices have a pronounced effect on soil chemical properties, such as nutrient content and balance, C/N ratio, pH and salinity (Gómez et al., 2011;Nieto et al., 2013). In the Mediterranean basin, olive trees (Olea europaea L.) and their cultivation have an important socio-economic role (Ferrise et al., 2013). Moreover, this type of culture can contribute to improve soil quality by favouring carbon sequestration in soils and limiting erosion depending on the practices applied i.e. natural grass but natural vegeta-tion or use of cover-crop (Hernández et al., 2005;Marquez-Garcia et al.,
2013).
SOM changes depending on various environmental factors can be successfully monitored by non-destructive techniques. FTIR spectros-copy is a rapid and non-destructive method well recognized for its abil-ity to characterize the chemical composition of living plants (Fourty
et al., 1996;Richardson et al., 2004) and plant litter (Farnet-Da Silva
et al., 2017). Authors have shown that this technique discriminated
soil depending on the vegetation cover i.e. grassland or forests (Ertlen
et al., 2015;Vysloužilová et al., 2015), alternative rotations treatments
or native prairie soil (Calderón et al., 2011). The impact of the drastic en-vironmental conditions on soil properties was also revealed through Near and Mid-infrared spectral profiles (or a both combination) from semi-arid Mediterranean soils with different management practices
(Aranda et al., 2014) or from insular Mediterranean soils (D'Acqui
et al., 2010). Based on Near and Mid-infrared spectroscopic data, Partial
Least-Square Regression (PLSR) can rapidly discriminate soils depend-ing on physical, chemical and microbial properties (total organic C, total N, CaCO3, microbial biomass…), which varied with above ground
vegetation assemblages (Janik et al., 2007;Knox et al., 2015) or with land use change (Paz-Kagan et al., 2014;Madhavan et al., 2017).
However, to our knowledge, no study has compared the relative ef-fects of different land uses on SOM profiles especially under Mediterra-nean climate conditions. Moreover, limited information is available about the effect of land use in interaction with climate conditions and with the practices associated to specific soil managements. Here, we aimed at identifying how land use (forest stands of Quercus spp. and Pinus spp. vs olive-tree orchards) shapes soil chemical signature. More-over, we tested whether this relation depended on applied practices (monospecific or mixed stands for forests, tillage, natural grass or olive-mill waste (OMW) amendments for olive-tree orchards) and on the contrasted climate conditions of the Mediterranean area at various spatial scales (humid vs sub-humid Mediterranean climate, coastal and inland areas from the sub-humid Mediterranean climate, Southern vs Northern slopes from the humid Mediterranean climate).
Based on Partial Least-Square Regression (PLSR), our research will eventually provide a model relying on the interactions between climate conditions, land use and soil chemical properties. This model could be further used to better assess the effects of these environmental factors on ecosystem functioning since soil molecular composition is a major driver regulating microbial activities involved in C dynamics (i.e. either sequestration or mineralization) in soils.
2. Materials and methods 2.1. Study sites and sampling design
The study was conducted in spring 2013 in South Eastern France (Provence-Alpes Côte d'Azur), an area characterized by frequent and in-tense droughts and heat waves typical of the Mediterranean climate. The Provence's soil is characterized by carbonatic pedofeatures, such asfine calcareous silt clay loam (Luvisol in the inland area according to theIUSS Working Group (2006)).
We conducted our study in two distinct bioclimates i.e. sub-humid and humid (Table 1), corresponding to altitudinal arrangement of veg-etation type i.e. Meso Mediterranean and Supra Mediterranean levels
(Quézel and Médail, 2003). Sampling plots were selected in i) a
500 km2area from Sausset les Pins to Port de Bouc (43° 22′N, 5°3′E)
for the sub-humid climate in coastal area (area 1), ii) a 700 km2area from the Massif de la Sainte Baume (43°21′ N, 5°43′E) to the Massif des Alpilles for the sub-humid climate in inland area (area 2) and iii) in a 700 km2area from the Massif of Luberon to the Réserve Géologique
of Digne les Bains (44°12′ N, 5°59′ E) for the humid climate (area 3). Dif-ferent annual trends and mean monthly temperature and precipitation between these areas are shown inTable 1.
In area 1,five independent 20-year-old olive tree orchards for each practice (tillage, natural grass, OMW amendment) were selected with the same variety of olive trees. In each orchard, a sampling plot (400 m2) was delimited and for each plot, 10 bulk soil samples (10 × 10 × 10 cm) were cored at 1 m of olive tree trunk using aflat shovel. The ten samples of soils were combined to obtain a unique composite sample per plot.
In area 2, the same experimental design was used for soil sampling in olive tree orchards.
Forest stands of around 60 ± 10-year-old, which had not undergone forest management for 35 years were selected: 30 square plots (20 m × 20 m) were chosen using GIS software (MapInfo professional 11.5 ®) and combining the French Geological Survey with the Forest Prop-erty Regional Center and the National Institute of Geography databases. For each of the three types of forest stand (Pinus halepensis, Quercus ilex and mixed stands 50/50),five square plots were selected on south-facing and north-south-facing slopes. Mixed stands (50/50) Pinus/Quercus were selected by measuring the basal area (m2.ha−1) of Quercus spp.
and Pinus spp. of each sampling plot, determined from each tree diam-eter measured at 1.20 m from the ground. Within each plot, ten bulk soil samples were collected from 0 to 10 cm deep (A horizon) and com-bined to obtain a unique composite sample per plot.
In area 3, the same protocol as described above was used for sam-pling in olive tree orchards and in forest stands of Pinus sylvestris, Quercus pubescens and mixed stands 50/50.
All the plots (in forest or orchards) were considered to be true repli-cates as the distance between them exceeded spatial dependence (N2 km).
The composite samples were transported to the laboratory at ambi-ent temperature, sieved to 2 mm and stored at 4 °C. These composite samples were dried at 70 °C for 72 h and ground to obtain a powder using a bullet blender Retsch MM40 (Fisher Scientific, France).
2.2. Chemical characterization of soils by FTIR-ATR (Fourier-Transformed Infrared– Attenuated Total Reflectance)
Soil samples were directly deposited on the attenuated total re flec-tance (ATR) accessory (Bruker“Golden Gate”) equipped with a dia-mond crystal prism (brazed in only one tungsten carbide part), four mirrors and two ZnSe focusing lenses in to reflect the optical path. Soil sample was pressed on the crystal area with the pressure arm and posi-tioned over the crystal/sample area. The recording of FTIR-ATR spectra was performed with a Thermo Nicolet IS10 spectrometer equipped with a MCT detector, an Ever-Glo source and a KBr/Ge beam-splitter, at room temperature. Data acquisition was done in absorbance mode from 4000 to 650 cm−1with a 4 cm−1nominal resolution. For each spec-trum, 100 scans were co-added. A background scan in air (in the same resolution and scanning conditions used for the samples) was carried out before the acquisition. The ATR crystal was carefully cleaned with ethanol to remove any residual trace of the previous sample. Three
spectra were recorded for each sample. Spectral zone between 4000 and 1800 cm−1was withdrawn since the spectral zone ranging 4000 and 3000 cm−1presents a broad band representative of O\\H, N\\H vi-brations with a lot of hydrogen bonding effects encompassing a part of aliphatic bands and the spectral region from 2400 to 1900 cm−1presents bands of CO2and diamond absorption. This is why this part of the
spec-trum is usually deleted in the analyses of the data. Multiplicative Scatter Correction (MSC) was used to compensate for additive and/or multipli-cative effects in spectral data for chemometrics treatments.
2.3. pH, elementary analysis and C-CaCO3content
Briefly, pH H2O was determined in distilled water after 45 min under
magnetic stirring at 600 rpm. Extractable phosphorus (P), was mea-sured with Olsen method and available potassium (K), iron (Fe) and cal-cium (Ca) byflame photometry. Total C and N contents were measured using a C/N elementary analyser (Flash EA 1112 serie ThermoScientific) on the dried cylindrical soil-core fractionb2 mm. Determination of cal-cium carbonate (CaCO3) in samples is based on the volumetric analysis
of the carbon dioxide CO2, which is released during the application of
hydrochloric acid solution HCl 4N and was performed using volumetric calcimeter Bernard modified method (Hulseman, 1966;Muller and
Gatsner, 1971). Carbon proportion of CaCO3was then calculated using
the molar mass:%C−CaCO3 ¼11100:991 %CaCO3.Then, organic carbon
was calculated by the difference between total C content and C-CaCO3
content (Wang et al., 2012). 2.4. Statistical analyses
The AComDim (Anova Common Dimension) method is a multi-block analysis highlighting the influential factors (and their interac-tions) by a simultaneous analysis of all data. This method is based on the same concept as ANOVA-PCA (also called APCA) and its description can be found inAmat et al. (2010)andBouveresse et al. (2011). AComDim decomposes the experimental data matrix into successive matrices and a multi-block PCA analysis of all matrices is performed in order to extract the“Common Components” (CCs) and their specific contribution, called salience, which make possible tofind the significant factors. The block significance is afterward estimated with a Fisher test (student Fisher F-test with n− 1 degrees of freedom, where n is the number of blocks and an alpha level equal to 0.01) applied on the
F-Table 1
Characterization of the two bioclimates and vegetation stages of the different areas under study (climate data from WorldClim dataset, 1950–2000,http://www.worldclim.org/,Hijmans et al., 2005).
Sampling area Area 1 Area 2 Area 3
Bioclimates Sub-humid Sub-humid Humid
Vegetation stages Meso-Mediterranean Meso-Mediterranean Supra-Mediterranean Location 43°22′ N, 5° 3′ E (coastal) 43°21′N, 5°43′E (inland) 44°12′N, 5°59′E
Mean altitude 30 m 400 m 1000 m
Main tree species Pinus halepensis Quercus
ilex Pinus halepensis Quercus pubescens Pinus sylvestris Associated vegetation Quercus coccifera, Pistacia lentiscus, Asparagus acutifolius Rosmarinus officinalis,
Cistus albidus
Asparagus acutifolius, Helichrysum stoecchas, Juniperus oxycedrus, Phillyrea latifolia, Quercus coccifera, Rhamnus alaternus, Rosmarinus officinalis, Rubia peregrina, Thymus vulgaris. Amelanchier ovalis, Aphyllanthes monspelliensis, Buxus sempervirens, Crataegus monogyna, Hieracium murorum, Juniperus communis, Rhamnus saxatilis, Sanguisorba minor, Teucrium polium
Annual precipitation 670 mm 755 mm 842 mm
Precipitation of the driest month 15 mm 18.2 mm 40.6 mm
Mean annual temperature 13.7 °C 13.00 °C 9.60 °C
Min temperature of the coldest month
values (Fi) calculated as
Fi¼λres
λi ð1Þ
whereλresis the salience of the residual block on CC1 andλiis the
sa-lience of the ithblock on CC1.
The blocks for which Fiis greater than the critical value (Fc) of the
Fisher table are considered as being related to influential factors or in-teractions. By examining the calculated saliences, it is possible to deter-mine which CC is related to which factor or interaction In order to estimate the effect of the factor, it is possible to plot the sample scores on the informative CCs vs. CC1. Loadings show spectral bands discrimi-nating samples for a CC and are used to understand sample projections on the score graphs(Gaston et al., 2017). All the spectral bands were assigned according toSocrates (2007). All computations were per-formed using Matlab 7.14 (R2012a). The AComDim procedure was adapted from the ComDim function in the free toolbox SAISIR
(Cordella and Bertrand, 2014).
Partial Least Square regression (Geladi and Kowalski, 1986;Van
Roon et al., 2014) is a concept of multivariate calibrations forms
where variables data are divided into groups of response variables and predictor variables. Here predictor variables were spectral profiles and responses were land use, climate or practices, binary codified. A linear regression equation was established where reflectance at each spectral wavelength (Xi) was related to a response variable (Y):
Yp¼ a1X1þ a2X2þ … þ anXnþ b ð2Þ
where Ypwas the predicted value for the response variable, a1to an
were the PLS regression coefficients for predictor variables (Xi) and b
was the equation intercept. The objective of this regression was to ex-tract“components” or latent variables (LV) responsible for the variation of the predictor variables and which best model the behavior of the re-sponse variables. The choice of extracted latent variables was done by cross-validation which consisted of constructing a calibration model on a subset of X data, then testing the model obtained on another subset of X data (not taken into account in the calibration model). A validation model was obtained for which the number of components was chosen so as to minimize the estimation error.
The quality of the calibration models was estimated by the standard error of calibration (SEC) (Eq.(3)) and the robustness of the model was evaluated by the standard error of prediction (SEP) (Eq.(4)).
SEC¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi PN i¼1 Yi−Y0i 2 N−1−p s ð3Þ SEP¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi PM i¼1 Ci−C0i 2 M s ð4Þ
where Yiis the measured value of the response variable and Yi′ the
cal-culated value from the calibration or validation equation. N corresponds to the number of samples used for the construction of the model, p to the number of optimal independent variables determined by cross-validation, and M to the number of samples used for the prediction. For the building of some PLS models, spectral data were averaged per sample.
The statistical significance of climate, land use and practices on soil chemical parameters was determined by three-way analysis of variance (ANOVA) and statistically significant (P b 0.05) main effects and interac-tions were analysed further using the Tuckey HSD post hoc tests. Data that were not normally distributed, or showing unequal variance were
transformed prior to analysis, using log10 transformation. Tab
le 2 Studied factors, le v el fa ctors and n um be r o f spectral d ata u se d for th e d ifferent Acom Dim stud ie s. Studied factors Land use Climate Practice Exposure Levels Olive-tree orchards Forests Sub-humid inland Sub-humid coastal Humid No-tillage Tillage
Olive-mill waste amendment Pinus halepensis Quercus ilex Pinus sylvestris Quercus pubescens Mixture South-facing North-facing First study 52 180 116 / 116 20 16 16 30 30 30 30 60 / / Second study 7 8 / 2 62 62 6 3 0 2 4 2 4 / / / / / / / Third study / 180 90 / 9 0 / / / 30 30 30 30 60 90 90 /: level n o t co ns idered .
3. Results
Wefirst tested the effects of land use (agrosystem vs forest), climate (sub-humid vs humid climate), the practices applied (agricultural, sylvicultural) and their interactions (first study) on soil chemical signa-ture. Then, for each land use, the effect of specific climate factors and practices were determined. For olive-tree orchards, the effects of cli-mate across a clicli-mate gradient (sub-humid coastal, sub-humid inland and humid climate), agricultural practices and their interactions were tested (second study). For forests, the effect of climate conditions at re-gional (sub-humid vs humid) and at local (Northern vs Southern slopes) scales, sylvicultural practices and their interactions were analysed (third study). For each study, the number of spectra for each factor is presented inTable 2.
3.1. Infrared spectroscopy reveals the effect of land use on soil chemical signature
To determine how land use‘fingerprint’ and/or practices applied modified soil chemical properties and whether it depended on climate conditions, FTIR-ATR technique was used to characterize the two sets of soils from either forests or agrosystems from the sub-humid and
humid climate (Table 2,first study). The F-values (computed from the saliences on CC1) of the different blocks (Fig. 1a) were close to 1 but only block 4 (“climate-land use interaction”) was larger than the critical F-value (Fc= 1.23) of the Fisher table according to the alpha level (α =
0.05 by considering 261 degrees of freedom). The significance of the three main factors (climate, land use and practice) was border line with F-values comprised between 1.15 and 1.20 for an alpha level of 0.05. The study of saliences confirmed the results obtained with the F-values (Fig. 1b). The effect of the interaction between climate and land use on FTIR-ATR spectra is shown by the projections associated to CC3 (related to block 4“interaction climate-land use”) inFig. 2a. The score plots of CC3 clearly shows a discrimination between soil samples de-pending on climate (humid or sub-humid) and land use (forests vs olive tree orchards). From the loading plot (Fig. 2b), the projections ac-cording to CC3 are positively explained by IR signals assigned toν(C=C) phenolic acid and (C\\H) out of plane (CH3) (Fig. 2b) i.e. 1630,
971 cm−1. This indicates that the chemical signature of forests soils from sub-humid climate and orchards from humid climate can be clearly differentiated by compounds from lignin, while soils of orchards from sub-humid climate and of forests from humid climate are charac-terized by the presence of carbonate, a key component of soils on calcar-eous parent material. These results are in accordance with those found
1 2 3 4 5 6 7 8 0 0.05 0.1 0.15 0.2 0.25 0.3 1 2 3 4 5 6 7 8 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Saliences on CC3 (1x2 interaction) Saliences on CC1 (residual noise)
(b) Saliences
(a) F-values
x e d n i k c o l B x e d n i k c o l B 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1 2 3 4 5 6 7 8 F-v a lu es Fc= 1.23 (a =0.05) Block indexFig. 1. AcomDim of spectral soil data from forest stands and olive-tree orchards: (a) F-values vs. block index, (b) saliences vs. block index for significant CCs. Block 1 “climate”, block 2 “land use”, block 3 “practices”, block 4 “climate-land use interaction”, block 5 “climate-practices interaction”, block 6 “land use-practices interaction”, block 7 “climate-land use-practices interaction”, block 8 “residual”.
with the other soil chemical properties analysed in this study. Higher amounts of C-CaCO3were found in soils of olive-tree orchards from
sub-humid climate and of forests from humid climate and moreover, these forests had lower amount of Organic-C (Table 3).
Though FTIR-ATR spectra revealed that projections were discrimi-nated for each type of land use the effect of practices appeared non-significant.
3.2. Infrared spectroscopy reveals the effect of practices applied for each land use
FTIR-ATR spectra were further analysed for each type of land use in-dependently (Table 2, second and third studies).
For soils from olive-tree orchards (Table 2second study), we thus tested the effect of climate (coastal and inland areas in sub-humid cli-mate and humid clicli-mate) and of the practices applied (tillage, natural grass and OMW amendment). AComDim analysis of spectral data of olive-tree orchards revealed that the critical F-value (Fc) was equal to 1.34. Under these conditions, only thefirst block, relative to the climate factor, had an F-value superior to Fc (plots of F-values and saliences ver-sus block index not shown). The score plots of CC1 verver-sus CC2 (Fig. 3a) showed that soils were clearly separated depending on climate. Projec-tions of soils from the humid climate were explained on the positive part of the loading on CC2 (Fig. 3b) by signals at 1633, 1072, 973, 796 and 777 cm−1assigned to (C=O) vibrations (in carboxylic acids/esters) or (C=C) stretching (in aromatics), (C\\H) out-of-plane deformation (CH3) and (=C-H) out-of-plane deformation bands (in aromatics)
Sub-humid climate
Humid climate
Scores on CC3
1
C
C
no
ser
oc
S
Forest stands
Olive-tree orchards
-0.1
0
0.1
-0.3
-0.2
-0.1
0
0.1
1800
1600
1400
1200
1000
800
650
-0.8 0.2 0 -0.2 -0.4 -0.6Wavenumber (cm
-1
)
)
u
a(
yti
s
n
et
nI
1629
1403
871
971
71
1
846
b
a
respectively. Soils from the coastal sub-humid climate were explained on the negative part by signals at 1400, 871 and 711 cm−1, which can be assigned to the presence of CaCO3. Soils from the inland area of
sub-humid climate were centered on the origin of the two-dimensional plots (Fig. 3a). The effect of practices was non-significant using this set of data. On the other hand, when considering the coastal and inland areas from sub-humid climate, the effect of practices was re-vealed (Fig. 4). Thus, no discrimination of samples depending on the practices was actually possible when data from the humid climate were included in the analyses. The effect of practices was observed ac-cording to score plots of CC1 versus CC3: soils complemented with OMW amendment were discriminated from tilled and non-tilled soils
(Fig. 4a). Projections assigned to OMW amendment were linked to
pos-itive signals of loading on CC3, pointed at 1083, 970 and 900 cm−1linked to C\\O of various alcohol types, (=C-H) out-of-plane deformation bands (in aromatics). Soils under tillage and natural grass were ex-plained by signals at 1409, 871 and 711 cm−1from CaCO3. This has to
be linked with the amounts of organic C which were significantly higher in soils complemented with OMW than in soils under tillage or natural grass (F = 5.025, pb 0.05, data not shown).
For soils from forest stands (Table 2third study), we tested the effect of climate (sub-humid and humid), of the sylvicultural practices applied (monospecific stands of Q. ilex, Q. pubescens, P. halepensis or P. sylvestris, and mixed stands of Q. ilex/P. halepensis and Q. pubescens/P. sylvestris) and of exposure (Southern vs Northern slope). A critical F-value of 1.28 (α = 0.05 by considering 180° of freedom) was found from ACOMDIM analysis and consequently factors“climate” and “practice” and interaction“land use - exposure” showed a F-value superior to Fc
and can be considered as significant factors. The score plots of CC4 (cor-responding to block 1 related to climate) showed discrimination of soils according to climate (Fig. 5a). Associated loading (Fig. 5b) characterized soils from humid climate by positive signals at 1411, 1164, 1122, 1083, 796, 777 cm−1assigned to C_C (Aromatic), νa(C-O-C) ester, C\\H
(Ar-omatic),γ(=C-H) out of plane aromatic, C\\H out of plane (disubsti-tuted aromatic ring), (Table 1) from phenolic compounds and cutins and by signals at 1394, 871 and 711 cm−1assigned to CaCO3. These
re-sults are corroborated with the amount of CaCO3found in soils from
humid climates which were higher than those from sub-humid climate
(Table 3). The differentiation of samples from sub-humid climate was
linked to signals at 1564, 908, 887, 727 cm−1(Fig. 5b) on the negative part of CC4 loading, which can be assigned to C_C from aromatics, C\\H out of plane (monosubstituted aromatic ring), C\\H out of plane (disubstituted aromatic ring) from phenolic compounds.
The score plot of CC3 (corresponding to block 2 related to practices)
(Fig. 5c) showed that soils under Pinus spp. stands were gathered on the
positive part of CC3 and were explained by the positive bands of CC3 loading at 1411, 871 and 711 cm−1i.e. from CaCO3(Fig. 5d). On the
other hand, samples from Quercus spp. were projected on the negative part of CC3 scores plot and explained by negative signals at 1612, 977, 908, 794, 775, 744 cm−1related to v(C=C) aromatic or conjugated, γ(=C-H) out of plane aromatic, ρ(HOH) from aromatics. Projections from soils of both Quercus spp. monospecific stands overlapped, while those of both Pinus spp stands were not strictly superimposed. Soils from both mixed stands (Q. ilex/P. halepensis and Q. pubescens/P. sylvestris) were centered at the origin of the two-dimensional plots, showing an equitable influence of the two tree species on soil properties.
The effect of exposure (northern vs southern slope) on the chemical signature of forest soils was found according to CC8, corresponding to block 6 related to interaction“practice – exposure” (Fig. 5e). Soils from either P. sylvestris and Q. pubescens stands were particularly dis-criminated depending on exposure compared to soils from P. halepensis and Q. ilex stands. Moreover, stand admixture has smoothen the effect of exposure as shown by projections gathered at the center of the two-dimensional plots. Soils under northern exposure of both stands from humid climate were linked to signals at 1394, 871 and 711 cm−1 i.e. from CaCO3(Fig. 5f). Projections of South-exposed soils from the
same stands were explained by signals at 906, 889, 777 and 720 cm−1 assigned toγ(=C-H) out of plane aromatic, ρ(HOH) from aromatics andδ(CH2) rocking (long chains).
3.3. Prediction of the environmental context (climate and land use) of soils Regarding the ability to differentiate the two types of land use con-sidered, the quality of the PLS predictions using entire FTIR profiles and a binary codification to differentiate forest and olive-tree orchards was significant (Table 4). Calibration performed on averaged spectral data gave an R2value of 0.99 and a low SEC. Parameters of the validation model showed that land use can be predicted (SEP equal to 0.24). The first regression coefficient (data not shown) confirmed the discrimina-tion of the two land uses on the bases of specific spectral bands. These bands were pointed at 1400, 1162, 1087, 870 and 711 cm−1for olive-tree orchards and 3300, 1602, 973 and 908 cm−1for forests, characteriz-ing the presence of carbonates or aromatic compounds respectively. Here, the implication of carbonate content in olive-tree orchards has a positive effect on the accuracy of PLS models to differentiate land use.
Table 3
Averages of soil chemical properties from forest soils (n = 30) and olive-tree orchards (n = 15) and p values from one-way ANOVA. Means with the same letter are not significantly dif-ferent (Tukey's Test).
p-Value Sub-humid coastal climate Sub-humid inland climate Humid climate Forest soils %N b0.001 – 0.52 ± 0.10 a 0.39 ± 0.09 b Organic C (%) b0.005 – 9.90 ± 1.68 a 6.88 ± 0.89 b C-CaCO3(%) b0.05 – 19.86 ± 5.81 a 28.84 ± 4.6 b C/N b0.05 – 25.83 ± 2.33 a 28.92 ± 3.21 b Ca (g.kg−1) b0.05 – 13.2 ± 3.1 a 16.4 ± 4.1 b P Olsen (mg.kg−1) ns – 14 ± 2 a 15 ± 4 a K (mg.kg−1) ns – 231 ± 81 a 203 ± 66 a Fe (mg.kg−1) ns – 1.1 ± 0.2 a 1.8 ± 0.5 a
Olive-tree orchard soils
%N ns 0.17 ± 0.02 a 0.19 ± 0.04 a 0.19 ± 0.04 a Organic C (%) ns 2.80 ± 0.88 a 3.19 ± 0.80 a 2.59 ± 0.49 a C-CaCO3(%) b0.05 27.88 ± 5.23 ab 35.20 ± 4.86 a 19.15 ± 4.11b C/N ns 17.51 ± 6.24 a 17.05 ± 4.81 a 13.31 ± 0.81 a Ca (g.kg−1) b0.05 15.6 ± 2.2 ab 19.1 ± 1.1 a 12.8 ± 2.5 ab P Olsen (mg.kg−1) ns 8 ± 1 11 ± 3 10 ± 4 K (mg.kg−1) ns 148 ± 78 a 202 ± 88 a 185 ± 95 a Fe (mg.kg−1) ns 2.1 ± 0.3 a 2.8 ± 0.5 a 1.8 ± 0.5 a
For climate prediction, cross-validation PLS resulted in a standard error of 0.28 and an R2equal to 0.86 and can thus be considered as significant.
Thefirst regression coefficient brought out the same spectral bands pre-viously identified to discriminate sub-humid (3259, 1371, 870 and 711 cm−1) from humid climate (1083, 796, 777 and 700 cm−1). PLS cross validation performance was poorly improved by separating data per land use and without spectral data average.
4. Discussion
To our knowledge, few studies aimed at deciphering how the chem-ical signature of SOM differed depending on factors acting at various spatial scales. Here, we tested the effects of land use and of the associ-ated practices under contrasted climate conditions at regional (sub-humid vs (sub-humid climate, coastal vs inland area) and local (southern vs
northern slopes) spatial scales. Though previous studies proposed models to predict SOM content and composition under agricultural or forested land use (Janik et al., 2007;Madhavan et al., 2017), authors underlined that specific calibrations should be realised considering main drivers of soil chemical signature such as climate and practices.
4.1. Land use shapes soil chemical signature depending on climate condi-tions as revealed by FTIR–ATR
This study clearly revealed a discrimination of soils under different land uses based on the chemical signature of both organic matter and mineral fractions of soil using FTIR technique. It has to be mentioned that CaCO3was found as the part of the mineral fraction which was
in-volved in the discrimination of soils. This goes in line with the mineral composition of the calcareous soils studied here, with very low
1800
1600
1400
1200
1000
800
650
0.3
0
-0.3
-0.6
1400
871
846
796 777
71
1
Wavenumber (cm
-1
)
)
ua
(
yti
s
ne
t
nI
1633
1072
973
a
b
-0.2
-0.1
0
0.1
0.2
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
Scores on CC2
Humid climate
Inland
Sub-humid climate
Olive-mill waste amendment
Tillage
No-tillage
1
C
C
n
o
se
r
oc
S
Coastal
Sub-humid climate
concentrations in iron and potassium compared to calcium (104fold
less) linked to CaCO3. Interestingly, this differentiation depended on
the bioclimate considered: forest soils from humid climate and orchard soils from sub-humid climate were characterized by CaCO3. This is
cor-roborated by the higher CaCO3amounts found in these soils. These
re-sults can be explained depending on the ecosystem considered. For forest stands, this is in accordance with our previous study:Brunel
et al. (2017)found that concentrations in CaCO3in forest soils of P.
sylvestris and Q. pubescens (from Humid climate) were higher than those from P. halepensis and Q. ilex (Sub-Humid climate). For
olive-tree orchards, a soil chemical signature linked to CaCO3may reveal a
poor soil quality.Comino et al. (2018), indeed showed that in agricul-tural soils with low organic matter, contribution of the mineral fraction to FTIR spectra was strong. Consequently, they discriminated soils richer in SOM but with low carbonate concentrations from those with oppo-site trends considering these chemical characteristics. On the other hand, forest soils from sub-humid climate and orchard soils from humid climate were characterized by aromatics. Several previous stud-ies underlined the specific effects of tree species on forest soil properties through above- and below-ground soil inputs, which modified organic
Olive-mill waste amendment
Tillage
No-tillage
Scores on CC3
1
C
C
no
ser
oc
S
-0.2
-0.1
0
0.1
0.2
0.3
-0.4
-0.2
0
0.2
1409
970
900
1083
871
71
1
846
-0.3
-0.2
-0.1
0
0.1
1800
1600
1400
1200
1000
800
650
Wavenumber (cm
-1
)
)
u
a(
yti
s
n
et
nI
a
b
matter quality and quantity (Iovieno et al., 2010;Kara et al., 2008). Q. ilex is supposed to exhibit higher aromatic contents than Q. pubescens as described in our previous study using solid-state NMR of13C to
qualify organic matter of Mediterranean tree species (Farnet et al., 2013). This also explained the differences in the chemical signature of forest soils depending on the climate considered (sub-humid vs humid climate). Concerning olive-tree orchards, the more favourable climate conditions in the humid climate may have limited the effect of conventional olive-grove agricultural practices, which often led to ero-sion producing a severe degradation of soil fertility (García-Ruiz et al., 2009). Here the aromatic signature found in olive tree orchards of humid climate can be linked to humification and organic matter aggre-gation that occurred in soils. Thus, FTIR showed that a soil chemical sig-nature assigned to a land use also depended on climate revealing that climate conditions were another important factor structuring SOM.
Aranda et al. (2014)also successfully differentiated soils from olive
groves and forests of Quercus spp. in the Mediterranean area, though these authors did not consider the geographical location as a driving fac-tor of soil chemical signature. It would be of huge importance to deter-mine, under the Mediterranean context, whether this chemical differentiation is preserved as a chemicalfingerprint “legacy” in the
Table 4
PLS results for the prediction of land use and climate factors (considering soils from forest and olive-tree orchard together or separately).
Calibration Validation Nb R2
SEC LV Nb R2
SEV Land use (forest and olive-tree orchard
data)a
58 0.99 0.07 15 29 0.88 0.23
Climate (forest and olive-tree orchard data)a
58 0.87 0.24 5 29 0.86 0.28
Climate (olive-tree orchard data)b
57 0.89 0.16 3 24 0.79 0.24 Climate (forest data)b
120 0.97 0.12 13 61 0.79 0.43
a
Averaged data.
b
Not averaged data.
1800 1600 1400 1200 1000 800 650 -0.3 0.3 0.2 0.1 -0.1 -0.2 0 1564 141 1 1394 1 164 11 2 2 1083 871 796 777 71 1 694 887 908 727 Wavenumber (cm-1) ) u a( yti s n et nI 1800 1600 1400 1200 1000 800 650 -0.2 0.5 0.3 0.2 0.1 0 -0.1 0.4 141 1 871 908 71 1 1612 977 794 775 744 Wavenumber (cm-1) ) u a( yt i s n et nI -01 0 0.1 0.2 0.3 1800 1600 1400 1200 1000 800 650 Wavenumber (cm-1) In te n si ty (a u ) 1394 871 71 1 777 906 889 846 1616
a
b
c
d
e
f
-0.2 -0.1 0 0.1 0.2 -0.2 -0.1 0 0.1 0.2 Scores on CC41
C
C
no
ser
oc
S
Humid climate Sub-humid climate -0.2 -0.1 0 0.1 0.2 -0.2 -0.1 0 0.1 0.2 Scores on CC31
C
C
no
ser
oc
S
QI Mixture PH QB PS-0.2
-0.1
0
0.1
0.2
-0.2
-0.1
0
0.1
0.2
Scores on CC81
C
C
no
se
ro
c
S
PS-S PS-N QB-S QB-N PH-S PH-N M-N M-S QI-N QI-SFig. 5. AcomDim results on spectral soil data from forest stands: (a) CC1 vs. CC4 scores plots, (b) loading on CC4 corresponding to factor“climate”, (c) CC1 vs. CC3 scores plots, (d) loading on CC3 corresponding to factor“land use”, (e) CC1 vs. CC8 scores plots, (f) loading on CC8 corresponding to interaction “land use- exposure”.
deeper layers of soils since the history of land use can impact present soil structure and functioning (Oyonarte et al., 2008). Recently,Ertlen
et al. (2015)have taken thefirst steps in deciphering the soil chemical
signature of past land uses and the associated vegetation using NIR in soils from North Eastern France.
It has to be noted that the mineral fraction, here mainly assigned to CaCO3, is key to define soil chemical signature.Comino et al. (2018)
in-dicated that the proportion of organic matter in soil can be particularly weak in agricultural soils and consequently mineral fraction is an im-portant part of the origin of spectral variation. In bulk soils, carbonate can be found under different forms as calcite (CaCO3) and/or dolomite
(CaMg (CO3)2). Both calcite and dolomite can be present in different
ratio depending on soil formation and geological background. FTIR pro-files of pure calcite and dolomite showed few differences as described
by Bruckman and Wriessnig (2013). Gunasekaran and Anbalagan
(2006)confirmed that calcite and dolomite groups were characterized
by infrared reflection band at 712 and 728 cm−1respectively. In our
case, AcomDim revealed the presence of calcite through specific peaks at 1400 cm−1(νasCO3), 871 cm−1(γasCO3) and 711 cm−1(δsCO3).
4.2. Practices are determinant in defining soil chemical signature and it de-pends on climate conditions at various spatial scales
Considering soils from olive-tree orchards located in the sub-humid climate, our study revealed that FTIR spectra showed variations in the chemical signature according to the practices applied (tillage and natu-ral grass vs OMW amendment).Boukhdoud et al. (2016)found that the effects of agricultural practices on soil properties were actually more pronounced in olive-tree orchards under more drastic climate condi-tions: the geographical location (coastal vs inland area) affected how agricultural practices influenced soil properties and the authors con-cluded that soil management should take into account certain specific environmental conditions. Here, OMW amendment, compared to tillage or no-tillage, provoked a change in FTIR spectra, revealing a chemical signature linked to aromatics. The study ofAranda et al. (2016), which analysed the long-term effect of OMW amendment on soil hydropho-bicity and wettability, indicated that, though no change in mineralogy was observed, olive pomace addition induced an increase in organic car-bon. Discriminating soils depending on the practices applied is a partic-ularly important result since most studies using FTIR actually showed the effect of land use on soil chemical signature but few revealed the fin-gerprint of the practices applied. For instance,Comino et al. (2018)did notfind any significant differences in olive grove soils under conven-tional or organic practices, though convenconven-tional management included tillage (at 20 cm depth, 3 to 5 times a year) and the use of herbicides and mineral fertilizers, which are known to have a great impact on soil properties.
In our study, soils from different forest stands were discriminated depending on the functional traits of tree species (coniferous vs broadleaved and evergreen species) and on their admixture.Traversa
et al. (2008)successfully distinguished the plant coveringfingerprint
(from typical Mediterranean plant species from forests and shrublands) on dissolved organic matter in soils using FTIR. Tree species identity strongly modifies, via litter quality (C/N, lignin and tannin concentra-tions), soil microbial and chemical properties and thus the global rate of decay (Laganière et al., 2009;Moukoumi et al., 2006). Our study indi-cated that forest composition, driven by sylvicultural practices, had a great impact on soil chemical properties as shown before (Weand
et al., 2010) and that this can be efficiently described using FTIR-ATR.
Northern or Southern exposure appeared to be determinant in de-fining soil chemical signature and it was particularly true for stands in humid climate. In a previous study,Pailler et al. (2014)examined the relative effects of climate conditions and tree species on soil properties in South-Eastern France and found that the‘climatic-geographic factor’, defined by different variables i.e. distance from the sea, elevation, expo-sure, summer rainfall and mean annual temperature, together with tree
species strongly modified microbial catabolism. These modifications can consequently impact soil chemical signature at local scales as sug-gested here.
4.3. FTIR-ATR is a useful tool to predict the environmental context (climate and land use) of soils
The PLS prediction successfully distinguished soils depending on land use and climate under the Mediterranean context. This rapid and discriminating method could thus help provide referential data to fur-ther build model to forecast land use legacy (Ertlen et al., 2015) from deeper horizons of soils which is particularly relevant in the Mediterra-nean area. The surface and composition of European MediterraMediterra-nean for-ests were indeed strongly structured by wildfires and human activities leading to agro-sylvo-pastoral systems during the last century (Quézel
and Médail, 2003). From the 1930s, abandonment of land previously
de-voted to agriculture promoted massive colonization by resinous species such as Pinus halepensis Mill. and Pinus sylvestris (Tatoni and Roche, 1994). Thus, forests of different historical continuity, which probably leads to different soil properties, can be found in Mediterranean area, and models built from FTIR data could help at integrating land use leg-acy to understand soil functioning.
5. Conclusion
This paper underlined that soil chemical signature from agrosystems or forests can be discriminated by mineral (CaCO3) vs organic fractions
(aromatics) of soil. Moreover, the effect of practices on soil chemical sig-nature was also highlighted: in agrosystems, aromatics were linked to OMW amendment, in forests, CaCO3characterized Pinus spp. stands
and aromatics, Quercus spp. stands. All these subtle variations depended on Mediterranean climate conditions at different spatial scales and were successfully revealed by FTIR. Thus, this technique combined with che-mometric treatments (AcomDim and PLS methods) can easily discrim-inate soil identities depending on various environmental factors and predict land use depending on climate conditions. This preliminary study will be completed with other techniques such as X-rays or UV andfluorimetry to better explore soil spectral signature depending on climate and land use.
Acknowledgements
We are very grateful to Marjorie Sweetko for improving the English of the manuscript. The project wasfinancially supported by the French Environment and Energy Management Agency (ADEME) and the Pro-vence Alpes Côte d'Azur Region (France) and by a PhD grant from the Agency for Environment and Energy Management' and 'National Coun-cil for Scientific Research, Lebanon.National Council for Scientific Re-search in Lebanon (CNRSL). We are very grateful to the Forest Property Regional Center (CRPF) for their contribution and especially to Mr. Olivier Martineau for his valuable recommendations in determin-ing the selected sites. We would like to thank the four anonymous re-viewers whose remarks greatly improved the manuscript.
References
Amat, S., Dupuy, N., Kister, J., Rutledge, D.N., 2010.Development of near infrared sensors: detection of influential factors by the AComDim method. Anal. Chim. Acta 675, 16–23.
Aranda, V., Ayora-Cañada, M.J., Domínguez-Vidal, A., Martín-García, J.M., Calero, J., Delgado, R., Verdejo, T., González-Vila, F.J., 2011.Effect of soil type and management (organic vs. conventional) on soil organic matter quality in olive groves in a semi-arid environment in Sierra Mágina Natural Park (S Spain). Geoderma 164, 54–63.
Aranda, V., Domínguez-Vidal, A., Comino, F., Calero, J., Ayora-Canada, M.J., 2014. Agro-environmental characterization of semi-arid Mediterranean soils using NIR reflection and mid-IR-attenuated total reflection spectroscopies. Vib. Spectrosc. 74, 88–97.
Aranda, V., Calero, J., Plaza, I., Ontiveros-Ortega, A., 2016.Long-term effects of olivemill pomace co-compost on wettability and soil quality in olive groves. Geoderma 267, 185–195.
Boukhdoud, N., Gros, R., Darwish, T., Farnet Da Silva, A.M., 2016.Effect of agricultural practices and coastal constraints on soil microbial functional properties in Mediterra-nean olive orchards. Eur. J. Soil Sci. 67, 470–477.
Bouveresse, D.J.R., Pinto, R.C., Schmidtke, L.M., Locquet, N., Rutledge, D.N., 2011. Identifica-tion of significant factors by an extension of ANOVA-PCA based on multi-block anal-ysis. Chemom. Intell. Lab. Syst. 106, 173–182.
Bruckman, V.J., Wriessnig, K., 2013.Improved soil carbonate determination by FT-IR and X-ray analysis. Environ. Chem. Lett. 11, 65–70.
Brunel, C., Gros, R., Ziarelli, F., Farnet Da Silva, A.M., 2017.Additive or non-additive effect of mixing oak in pine stands on soil properties depends on the tree species in Medi-terranean forests. Sci. Total Environ. 590, 676–685.
Calderón, F.J., Mikha, M.M., Vigil, M.F., Nielsen, D.C., Benjamin, J.G., Reeves, J.B., 2011.
Diffuse-reflectance mid-infrared spectral properties of soils under alternative crop rotations in a semi-arid climate. Commun. Soil Sci. Plant Anal. 42, 2143–2159.
Comino, F., Aranda, V., García-Ruiz, R., Ayora-Cañada, M.J., Domínguez-Vidal, A., 2018. In-frared spectroscopy as a tool for the assessment of soil biological quality in agricul-tural soils under contrasting management practices. Ecol. Indic. 87, 117–126.
Cordella, C.B.Y., Bertrand, D., 2014.SAISIR: a new general chemometric toolbox. Trends Anal. Chem. 54, 75–82.
Coûteaux, M.M., Bottner, P., Berg, B., 1995.Litter decomposition, climate and litter quality. Tree 10, 63–66.
D'Acqui, L.P., Pucci, A., Janik, L.J., 2010.Soil properties prediction of western Mediterra-nean islands with similar climatic environments by means of mid-infrared diffuse re-flectance spectroscopy. Eur. J. Soil Sci. 61, 865–876.
Dhillon, D.S., Gillespie, A., Peak, D., Van Rees, K.J.C., 2017.Spectroscopic investigation of soil organic matter composition for shelterbelt agroforestry systems. Geoderma 298, 1–13.
Ertlen, D., Schwartz, D., Brunet, D., Trendel, J.M., Adam, P., Schaeffer, P., 2015.Qualitative near infrared spectroscopy, a new tool to recognize past vegetation signature in soil organic matter. Soil Biol. Biochem. 82, 127–134.
Farnet, A.M., Qasemian, L., Gil, G., Ferré, E., 2013.The importance of water availability in the reaction equilibrium of hydrolases in forest litters from a Mediterranean area: a study on lipases. Eur. J. Soil Sci. 64, 1–6.
Farnet, A.M., Boukhdoud, N., Gros, R., 2016.Distance from the sea as a driving force of mi-crobial communities under water potential stresses in litters of two typical Mediter-ranean plant species. Geoderma 269, 1–9.
Farnet-Da Silva, A.M., Ferré, E., Dupuy, N., De La Boussinière, A., Rébufa, C., 2017.Infra-red spectroscopy reveals chemical interactions driving water availability for enzyme ac-tivities in litters of typical Mediterranean tree species. Soil Biol. Biochem. 114, 72–81.
Ferrise, R., Moriondo, M., Trombi, G., Miglietta, F., Bindi, M., 2013.Climate change impacts on typical Mediterranean crops and evaluation of adaptation strategies to cope with. Regional Assessment of Climate Change in the Mediterranean. Springer, pp. 49–70.
Fourty, Th., Baret, F., Jacquemoud, S., Schmuck, G., Verdebout, J., 1996.Leaf optical proper-ties with explicit description of its biochemical composition: direct and inverse prob-lems. Remote Sens. Environ. 56, 104–117.
Garcia-Palacios, P., Shaw, E.A., Wall, D.H., Hättenschwiler, S., 2016.Temporal dynamics of biotic and abiotic drivers of litter decomposition. Ecol. Lett. 19, 554–563.
García-Ruiz, R., Ochoa, V., Viñegla, B., Hinojosa, M.B., Peña-Santiago, R., Liébanas, G., Linares, J.C., Carreira, J.A., 2009.Soil enzymes, nematode community and selected physico-chemical properties as soil quality indicators in organic and conventional olive oil farming: influence of seasonality and site features. Appl. Soil Ecol. 41, 305–314.
Gaston, F., Dupuy, N., Marque, S.R.A., Barbaroux, M., Dorey, S., 2017.Impact of γ-irradiation, ageing and their interactions on multilayerfilms followed by AComDim. Anal. Chim. Acta 981, 11–23.
Geladi, P., Kowalski, B., 1986.Partial least squares regression: a tutorial. Anal. Chim. Acta 185, 1–17.
Gómez, J., Llewellyn, C., Basch, G., Sutton, P., Dyson, J., Jones, C., 2011.The effects of cover crops and conventional tillage on soil and runoff loss in vineyards and olive groves in several Mediterranean countries. Soil Use Manag. 27, 502–514.
Gunasekaran, S., G. Anbalagan, G., Pandi, S., 2006. Raman and infrared spectra of carbon-ates of calcite structure. J. Raman Spectrosc. 37, 892–899.
Hernández, A., Lacasta, C., Pastor, J., 2005.Effects of different management practices on soil conservation and soil water in a rainfed olive orchard. Agric. Water Manag. 77, 232–248.
Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A., 2005.Very high resolution in-terpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965–1978.
Hulseman, J., 1966.An inventory of marine carbonate materials. J. Sediment. Petrol. 36, 622–625.
Iovieno, P., Alfani, A., Bååth, E., 2010.Soil microbial community structure and biomass as affected by Pinus pinea plantation in two Mediterranean areas. Appl. Soil Ecol. 45, 56–63.
IUSS Working Group WRB, 2006.World Reference Base for Soil Resources 2006.. World Soil Resources Reports NO. 103. FAO, Rome, Italy.
Janik, L.J., Skjemstad, J.O., Shepherd, K.D., Spouncer, L.R., 2007.The prediction of soil car-bon fractions using mid-infrared-partial least square analysis. Aust. J. Soil Res. 45, 73–81.
Kara, Ö., Bolat,İ., Çakıroğlu, K., Öztürk, M., 2008.Plant canopy effects on litter accumula-tion and soil microbial biomass in two temperate forests. Biol. Fertil. Soils 45, 193–198.
Kleber, M., Eusterhues, K., Keiluwet, M., Mikutta, C., Mikutta, R., Nico, P.S., 2015. Mineral-organic associations: formation, properties, and relevance in soil environments. Adv. Agron. 130, 1–140.
Knox, N.M., Grunwald, S., McDowell, M.L., Bruland, G.L., Myers, D.B., Harris, W.G., 2015.
Modelling soil carbon fractions with visible near-infrared (VNIR) and mid-infrared (MIR) spectroscopy. Geoderma 239–240, 229–239.
Laganière, J., Paré, D., Bradley, R.L., 2009.Linking the abundance of aspen with soil faunal communities and rates of belowground processes within single stands of mixed aspen–black spruce. Appl. Soil Ecol. 41, 19–28.
Liu, C., Lia, Z., Berhe, A.A., Xiao, H., Liu, L., Wang, D., Peng, H., Zeng, G., 2019.Characterizing dissolved organic matter in eroded sediments from a loess hilly catchment using fluorescence EEM-PARAFAC and UV–visible absorption: insights from source identifi-cation and carbon cycling. Geoderma 334, 37–48.
Madhavan, D.B., Baldock, J.A., Read, Z.J., Murphy, S.C., Cunningham, S.C., Perring, M.P., Herrmann, T., Lewis, T., Cavagnaro, T.R., England, J.R., Paul, K.I., Weston, C.J., Baker, T.G., 2017.Rapid prediction of particulate, humus and resistant fractions of soil or-ganic carbon in reforested lands using infrared spectroscopy. J. Environ. Manag. 193, 290–299.
Marquez-Garcia, F., González-Sánchez, E., Castro-Garcia, S., Ordóñez-Fernández, R., 2013.
Improvement of soil carbon sink by cover crops in olive orchards under semiarid con-ditions. Influence of the type of soil and weed. Span. J. Agric. Res. 11, 335–346.
Moukoumi, J., Munier Lamy, C., Berthelin, J., Ranger, J., 2006.Effect of tree species substi-tution on organic matter biodegradability and mineral nutrient availability in a tem-perate topsoil. Ann. For. Sci. 63, 763–771.
Muller, G., Gatsner, M., 1971.Chemical analysis. Neues Jb. Mineral. Monat. 10, 466–469.
Nieto, O., Castro, J., Fernández-Ondoño, E., 2013.Conventional tillage versus cover crops in relation to carbonfixation in Mediterranean olive cultivation. Plant Soil 365, 321–335.
Oyonarte, C., Aranda, V., Durante, P., 2008.Soil surface properties in Mediterranean mountain ecosystems: effects of environmental factors and implications of manage-ment. For. Ecol. Manag. 254, 156–165.
Pailler, A., Vennetier, M., Torre, F., Ripert, C., Guiral, D., 2014.Forest soil microbial func-tional patterns and response to a drought and warming event: key role of climate-plant-soil interactions at a regional scale. Soil Biol. Biochem. 70, 1–4.
Paz-Kagan, T., Shachak, M., Zaady, E., Karnieli, A., 2014.A spectral soil quality index (SSQI) for characterizing soil function in areas of changed land use. Geoderma 230–231, 171–184.
Polyakova, O., Billor, N., 2007.Impact of deciduous tree species on litterfall quality, de-composition rates and nutrient circulation in pine stands. For. Ecol. Manag. 253, 11–18.
Prandel, L.V., Saab, S.C., Brinatti, A.M., Giarola, N.F.B., Leite, W.C., Cassaro, F.A.M., 2014.
Mineralogical analysis of clays in hardsetting soil horizons, by X-rayfluorescence and X-ray diffraction using Rietveld method. Radiat. Phys. Chem. 95, 65–68.
Prescott, C.E., Grayston, S.J., 2013.Tree species influence on microbial communities in lit-ter and soil: current knowledge and research needs. For. Ecol. Manag. 309, 19–27.
Qasemian, L., Guiral, D., Farnet, A.M., 2014.How do microlocal environmental variations affect microbial activities of a Pinus halepensis litter in a Mediterranean coastal area? Sci. Total Environ. 496 (398–205).
Quézel, P., Médail, F., 2003.Ecologie et Biogéographie des Forêts Meéditerranéennes. En-vironmental SeriesElsevier, p. 571.
Richardson, A.D., Reeves III, J.B., Gregoire, G.T., 2004.Multivariate analyses of visible/near infrared (VIS/NIR) absorbance spectra reveal underlying spectral differences among dried, ground conifer needle samples from different growth environments. New Phytol. 161, 291–301.
Socrates, G., 2007.Infrared and RAMAN Characteristics Group Frequencies, Tables and Charts. 3rd ed. John Wiley & Sons Ltd.
Tatoni, T., Roche, P., 1994.Comparison of old-field and forest revegetation dynamics in Provence. J. Veg. Sci. 5, 295–302.
Traversa, A., D'Orazio, V., Senesi, N., 2008.Properties of dissolved organic matter in forest soils: influence of different plant covering. For. Ecol. Manag. 256, 2018–2028.
Van Roon, P., Zakizadeh, J., Chartier, S., 2014.Partial least squares tutorial for analyzing neuroimaging data. Tutor. Quant. Methods Psychol. 10, 200–215.
Vysloužilová, D., Ertlen, D.N., Šefrna, L., Novák, T., Virágh, K., Rué, M., Campaner, A., Dreslerová, D., Schwartz, D., 2015.Investigation of vegetation history of buried cher-nozem soils using near-infrared spectroscopy (NIRS). Quat. Int. 365, 203–211.
Wang, X.J., Wang, J.P., Zhang, J., 2012.Comparisons of three methods for organic and in-organic carbon in calcareous soils of Northwestern China. PLoS One 7, 44334.
Weand, M.P., Arthur, M.A., Lovett, G.M., McCulley, R.L., Weathers, K.C., 2010.Effects of tree species and N additions on forestfloor microbial communities and extracellular en-zyme activities. Soil Biol. Biochem. 42, 2161–2173.
